Different level of endogenous and IV

[Jenee Gu]

Dear All,
I hope this post finds you well. I want to use a county-level instrument variable on an individual-level endogenous variable, for example, y=b1x1*x2+b2x1+b2x2, x1 is an individual level endogenous variable, z1 is a county-level instrument, does this make sense?

Thank you!

[George Ford]

It is not uncommon. Main consideration is whether it meets the relevant criterion for a good IV.

William N. Evans, Wallace E. Oates, Robert M. Schwab, Measuring Peer Group Effects: A Study of Teenage Behavior, Journal of Political Economy, Vol. 100, No. 5 (Oct., 1992), pp. 966-991

Several citations of papers doing so are on page 2 https://www.nber.org/system/files/wo...12211.rev1.pdf